Identification of clinically actionable biomarkers via routine comprehensive genomic profiling across a large community health system

e15035

Background: Precision therapies and immunotherapies have revolutionized cancer care, with novel genomic biomarker-associated therapies being introduced into clinical practice rapidly. We assessed the utility of comprehensive genomic profiling (CGP)-based testing for identifying biomarkers associated with approved therapies, and therapies in precision medicine basket clinical trials (CT) across a large cohort of advanced cancer patients in the Providence health system. Methods: Advanced cancer patients were tested utilizing the Providence CGP workflow between 2019-2021. Clinical actionability was assessed for CGP and compared with results from an in silico 50-gene panel based on a previously utilized lab-developed test at Providence. Clinical actionability was assessed based on OncoKB and alterations were assigned as: FDA recognized (Level 1), standard of care (Level 2), FDA approved/investigational drug in another indication or having compelling clinical evidence (Level 3). CT matching was assessed based on enrollment criteria for ASCO-TAPUR, NCI-MATCH and My Pathway CTs at time of testing. Pooled electronic medical record and genomic data were curated and standardized. Results: Of the 3,218 advanced cancer patients tested with CGP, 52% were female, 80% were white, and median age was 67 years. Across 31 tumor types, the most commonly tested were lung (26%), bowel/colon (16%), and breast (9%). Overall, 48% of patients tested with CGP harbored at least one actionable biomarker (OncoKB Levels1/2/3). Clinical actionability was significantly higher in the CGP cohort compared to the in silico cohort based on presence of at least one Level 1 biomarker (45% vs. 19%, p < 0.001). CGP cohort had higher proportion of patients with multiple/co-occurring Level 1 biomarkers compared to in silico cohort (20% vs. 9%, p < 0.001). Of the most prevalent tumor types, 57% lung, 94% bowel/colon, and 37% breast had Level 1 alterations with CGP testing. Notably, 49% of CGP cohort vs. 23% in silico cohort (p < 0.001) harbored a biomarker matching to one or more arms of the three basket CTs. Conclusions: CGP and small panel testing can both identify patients eligible for approved therapies and/or basket CTs with CGP having significantly higher clinical actionability and CT eligibility. We expect value of CGP to increase as biomarker actionability transforms clinical practice.

Pathogenic fusion detection in solid malignancies utilizing RNA-DNA based comprehensive genomic profiling (CGP) testing

3078

Background: Gene fusions caused by chromosomal rearrangements comprise a key category of oncogenic driver mutations. However, given the diverse array of potentially novel loci where each proto-oncogene can translocate, many assays including DNA-based CGP have technical limitations that disallow the detection of all relevant fusion partners potentially leading to false negatives. Hybrid Capture RNA sequencing renders a more comprehensive evaluation of genes and allows detection of novel and known fusion partners. Here we assessed the impact of utilizing in-house CGP testing with a paired RNA-DNA hybrid assay in the identification of pathogenic fusions and their potential clinical actionability for patients with solid tumors across a large US health system. Methods: Patients in the Providence health system diagnosed with advanced solid tumor malignancies over a two-year period (2019-2021) received reflex CGP testing at the time of diagnosis utilizing an internally validated workflow. DNA/RNA sequencing results as well as histology and staging information were curated from deidentified electronic medical records and in-house databases, and tumor types were mapped to OncoTree tissue categories. Potential clinical actionability was assessed based on OncoKB therapeutic levels 1-3 and clinical trial eligibility matched to the biomarker inclusion criteria for ASCO TAPUR, NCI-MATCH and MyPathway studies (both without time limits and at time of testing). Results: The median patient age at diagnosis was 67 years, 52% of patients were female, and the majority (80%) were white. Across all tested advanced solid tumors, 6.7% (217/3218) were found to harbor a pathogenic fusion. The tumor types most enriched in this set of pathogenic fusions were prostate (30%), lung (27%), CUP (10%) and breast (9%). 29% (n = 64) of the identified pathogenic fusions were identified as actionable based on OncoKB criteria (levels 1-3), and 31% (n = 69) matched to one or more arms in the ASCO TAPUR, NCI-MATCH or MyPathway basket clinical trials. The most frequent actionable fusion driver genes identified were ALK (12%), FGFR 1-3 (12%), RET (7%) NTRK 1-3 (3%), and ROS1 (2%) and a subset of these key drivers were fused with novel gene pairs. A subset of fusions co-occurred with other targetable biomarkers, with the most common comprising tumor mutational burden high (TMB-H) (13%), PIK3CA (7%) and high microsatellite instability (MSI-H) (2%). Conclusions: In-house CGP testing utilizing an RNA-DNA based assay identified actionable fusion targets across tumor types, with many novel fusion partners that may be undetectable by prior generation sequencing assays. While many of these actionable targets are rare individually, the expanding totality of actionable gene alterations supports the growing utility of CGP for identifying patients who are candidates for approved targeted therapies and clinical trials.